Claude Sonnet 4 vs o3
Wins 0 of 7 areas
—
Wins 6 of 7 areas
Coding · Agents · Reasoning · Images and charts · Long documents · Following instructions
o3 is the stronger all-rounder.
Scores updated · 46 tests both models report · How we compare
Where each one wins
Tests won in each of the seven areas where both have results. Each piece is one test, so a longer bar means more evidence; grey means the two scored within a point of each other.
- ReasoningHard problems that need careful thinking03o33 of 5 tests · 2 ties
- Images and chartsUnderstanding pictures, charts and video03o33 of 3 tests
- AgentsCarrying out multi-step tasks on its own02o32 of 2 tests
- Following instructionsDoing exactly what it is asked02o32 of 2 tests
- CodingWriting and fixing software12o32 of 3 tests
- Long documentsFinding answers in very long texts01o31 of 1 test
- FactsGetting facts right instead of making them up11Even1 each
Long documents rests on a single test.
What it costs
Prices per million tokens, roughly 750,000 words. The bars show the cost of a million tokens read plus a million written.
o3 costs 44% less for the same work.
The biggest differences
The tests each model wins by the widest margin, up to three each. Scores are out of 100.
Where o3 pulls ahead
- Finds hard-to-locate facts by browsing the webBrowseComp+37.5points ahead
- Follows unfamiliar, precisely checkable instructionsIFBench+16.7points ahead
- Answers hard knowledge questions correctlyAA-Omniscience · Accuracy+15.9points ahead
Where Claude Sonnet 4 pulls ahead
- Avoids making up answers it doesn't knowAA-Omniscience · Non-hallucination+59points ahead
- Fixes real GitHub issues in Python projectsSWE-bench Verified+3.6points ahead
Every test, side by side
All 46 tests both models report. The winning score is in its model's colour; marks a score checked independently.
Codingo3
- Terminal-Bench Hardo3 by 631.137.1+6
- SWE-bench VerifiedClaude Sonnet 4 by 3.672.769.1+3.6
- SciCodeo3 by 14041+1
Reasoningo3
- Humanity's Last Examo3 by 9.410.720.1+9.4
- SimpleBencho3 by 7.645.553.1+7.6
- GPQA Diamondo3 by 577.782.7+5
- CritPttie0.31.1tie
- ARC-AGI-2tie5.96.5tie
FactsEven
- AA-Omniscience · Non-hallucinationClaude Sonnet 4 by 5970.911.9+59
- AA-Omniscience · Accuracyo3 by 15.922.738.6+15.9
Images and chartso3
- MMMUo3 by 8.574.482.9+8.5
- MMMU-Proo3 by 8.361.870.1+8.3
- LMArena · Visiono3 by 23 rating points11911214+23 rating
Following instructionso3
- IFBencho3 by 16.754.771.4+16.7
- Multi-Challengeo3 by 3.357.160.4+3.3
Other results28 tests, not counted
Tests outside the eight areas. They are not counted above: several are summary scores built from other tests, or the same test under another name.
- AIME 2024o3 by 48.243.491.6+48.2
- SimpleQAo3 by 33.515.949.4+33.5
- livecodebench_hardo3 by 33.432.666+33.4
- ZebraLogico3 by 22.173.795.8+22.1
- ARC-AGI-1o3 by 20.84060.8+20.8
- AA Agentic Indexo3 by 18.517.636.1+18.5
- LiveCodeBencho3 by 16.268.584.7+16.2
- τ²-Bench Telecom (AA run)o3 by 16.164.680.7+16.1
- AIME 2025o3 by 15.970.586.4+15.9
- AA-OmniscienceClaude Sonnet 4 by 15.70.2-15.6+15.7
- livecodebench_mediumo3 by 13.87689.8+13.8
- Aider-Polygloto3 by 10.670.781.3+10.6
- Tau2 airlineo3 by 9.355.564.8+9.3
- TAU-bench (airline)Claude Sonnet 4 by 86052+8
- TAU-bench (retail)Claude Sonnet 4 by 6.680.573.9+6.6
- τ²-Bench (Retail)o3 by 5.27580.2+5.2
- MATH-500 (EM)o3 by 4.19498.1+4.1
- AIR-Bench 2024Claude Sonnet 4 by 3.888.384.5+3.8
- frontiermath_tier_4_v1o3 by 2.102.1+2.1
- AA Intelligenceo3 by 1.318.920.2+1.3
- bbqo3 by 1.396.697.9+1.3
- MMLU-Proo3 by 1.383.785+1.3
- simple_safety_testsClaude Sonnet 4 by 110099+1
- anthropic_red_teamtie99.298.3tie
- Artificial Analysis Coding Indextie37.638.4tie
- XSTesttie96.697.3tie
- HarmBenchtie9898.4tie
- livecodebench_easytie98.899.1tie
Questions people ask
Which is better, Claude Sonnet 4 or o3?
o3 wins six of the seven areas where both have results: coding, agents, reasoning, images and charts, long documents and following instructions. Claude Sonnet 4 wins none. They are level on facts.
Which is better for coding?
o3. It wins 2 of the 3 coding tests both models report; Claude Sonnet 4 wins 1.
Which is cheaper?
Claude Sonnet 4 costs $3.00 per million input tokens and $15.00 per million output tokens; o3 costs $2.00 and $8.00. That makes o3 about 44% cheaper for the same work.
How do you compare the two?
We use the 46 benchmark tests both models have published scores on. The verdict counts the 18 tests in the eight capability areas, and a gap under one point (ten on rating-style scales) counts as a tie. The other 28 are listed but not counted, because several are summary scores or repeat a test. Each score is the one shown on the model's own page.